Computing explanations for interactive constraint-based systems

Constraint programming has emerged as a successful paradigm for modelling combinatorial problems arising from practical situations. In many of those situations, we are not provided with an immutable set of constraints. Instead, a user will modify his requirements, in an interactive fashion, until he is satisfied with a solution. Examples of such applications include, amongst others, model-based diagnosis, expert systems, product configurators. The system he interacts with must be able to assist him by showing the consequences of his requirements. Explanations are the ideal tool for providing this assistance. However, existing notions of explanations fail to provide sufficient information. We define new forms of explanations that aim to be more informative. Even if explanation generation is a very hard task, in the applications we consider, we must manage to provide a satisfactory level of interactivity and, therefore, we cannot afford long computational times. We introduce the concept of representative sets of relaxations, a compact set of relaxations that shows the user at least one way to satisfy each of his requirements and at least one way to relax them, and present an algorithm that efficiently computes such sets. We introduce the concept of most soluble relaxations, maximising the number of products they allow. We present algorithms to compute such relaxations in times compatible with interactivity, achieving this by indifferently making use of different types of compiled representations. We propose to generalise the concept of prime implicates to constraint problems with the concept of domain consequences, and suggest to generate them as a compilation strategy. This sets a new approach in compilation, and allows to address explanation-related queries in an efficient way. We define ordered automata to compactly represent large sets of domain consequences, in an orthogonal way from existing compilation techniques that represent large sets of solutions.

Effects of vehicle speed in structural health monitoring from operational conditions of bridges

The effects of vehicle speed for Structural Health Monitoring (SHM) of bridges under operational conditions are studied in this paper. The moving vehicle is modelled as a single degree oscillator traversing a damaged beam at a constant speed. The bridge is modelled as simply supported Euler-Bernoulli beam with a breathing crack. The breathing crack is treated as a nonlinear system with bilinear stiffness characteristics related to the opening and closing of crack. The unevenness of the bridge deck is modelled using road classification according to ISO 8606:1995(E). The stochastic description of the unevenness of the road surface is used as an aid to monitor the health of the structure in its operational condition. Numerical simulations are conducted considering the effects of changing vehicle speed with regards to cumulant based statistical damage detection parameters. The detection and calibration of damage at different levels is based on an algorithm dependent on responses of the damaged beam due to passages of the load. Possibilities of damage detection and calibration under benchmarked and non-benchmarked cases are considered. Sensitivity of calibration values is studied. The findings of this paper are important for establishing the expectations from different vehicle speeds on a bridge for damage detection purposes using bridge-vehicle interaction where the bridge does not need to be closed for monitoring. The identification of bunching of these speed ranges provides guidelines for using the methodology developed in the paper.

Early biomarkers to predict grade of encephalopathy following hypoxic ischaemic injury

The standard early markers for identifying and grading HIE severity, are not sufficient to ensure all children who would benefit from treatment are identified in a timely fashion. The aim of this thesis was to explore potential early biomarkers of HIE. Methods: To achieve this a cohort of infants with perinatal depression was prospectively recruited. All infants had cord blood samples drawn and biobanked, and were assessed with standardised neurological examination, and early continuous multi-channel EEG. Cord samples from a control cohort of healthy infants were used for comparison. Biomarkers studied included; multiple inflammatory proteins using multiplex assay; the metabolomics profile using LC/MS; and the miRNA profile using microarray. Results: Eighty five infants with perinatal depression were recruited. Analysis of inflammatory proteins consisted of exploratory analysis of 37 analytes conducted in a sub-population, followed by validation of all significantly altered analytes in the remaining population. IL-6 and IL-6 differed significantly in infants with a moderate/severely abnormal vs. a normal-mildly abnormal EEG in both cohorts (Exploratory: p=0.016, p=0.005: Validation: p=0.024, p=0.039; respectively). Metabolomic analysis demonstrated a perturbation in 29 metabolites. A Cross- validated Partial Least Square Discriminant Analysis model was developed, which accurately predicted HIE with an AUC of 0.92 (95% CI: 0.84-0.97). Analysis of the miRNA profile found 70 miRNA significantly altered between moderate/severely encephalopathic infants and controls. miRNA target prediction databases identified potential targets for the altered miRNA in pathways involved in cellular metabolism, cell cycle and apoptosis, cell signaling, and the inflammatory cascade. Conclusion: This thesis has demonstrated that the recruitment of a large cohortof asphyxiated infants, with cord blood carefully biobanked, and detailed early neurophysiological and clinical assessment recorded, is feasible. Additionally the results described, provide potential alternate and novel blood based biomarkers for the identification and assessment of HIE.

Cryptographic coprocessors for embedded systems

In the field of embedded systems design, coprocessors play an important role as a component to increase performance. Many embedded systems are built around a small General Purpose Processor (GPP). If the GPP cannot meet the performance requirements for a certain operation, a coprocessor can be included in the design. The GPP can then offload the computationally intensive operation to the coprocessor; thus increasing the performance of the overall system. A common application of coprocessors is the acceleration of cryptographic algorithms. The work presented in this thesis discusses coprocessor architectures for various cryptographic algorithms that are found in many cryptographic protocols. Their performance is then analysed on a Field Programmable Gate Array (FPGA) platform. Firstly, the acceleration of Elliptic Curve Cryptography (ECC) algorithms is investigated through the use of instruction set extension of a GPP. The performance of these algorithms in a full hardware implementation is then investigated, and an architecture for the acceleration the ECC based digital signature algorithm is developed. Hash functions are also an important component of a cryptographic system. The FPGA implementation of recent hash function designs from the SHA-3 competition are discussed and a fair comparison methodology for hash functions presented. Many cryptographic protocols involve the generation of random data, for keys or nonces. This requires a True Random Number Generator (TRNG) to be present in the system. Various TRNG designs are discussed and a secure implementation, including post-processing and failure detection, is introduced. Finally, a coprocessor for the acceleration of operations at the protocol level will be discussed, where, a novel aspect of the design is the secure method in which private-key data is handled